This week I finished the report on the radiometer data we collected. However I will be revising it with some new calculations soon. I have done a couple wavelength comparisons to look at patterns in the reflected spectra. The following table shows the calculation I have done.




Calculation               Mean         Standard Deviation

(550/500)                2.2789       0.2471

(550-500)               0.0619       0.0172

((550-500)/50)        1.2e-3       3.4e-4



The calculations in the above table only include the 12 light and dark radiometer readings and no data from the DiveSpec. DiveSpec data will be included after I add a couple lines to the script that performs these calculations. I will add all of this data to the paper next week. I was also considering fitting a parabola through some of reflectance spectra. In the DiveSpec data, and to a lesser extent in the dive light illuminated radiometer spectra, there is a nice curve that looks like the top of a parabola. I don’t know if this is common practice but a quadratic equation could be calculated between 500-675 based on the average of all the spectra. Then when looking at spectra of an unknown object, a quadratic best fit line could be run through 500-675. Then it can be identified as Didemnum or not, depending on how far this line deviates from the previous parabola we have calculated based on our field work. Just for reference look at the following graph.

From wavelengths 500 to 675 the spectra looks like a nice parabola. This curve isn’t so obvious in the Satlantic radiometer data, but it may be worth a shot…

I have been putting the finishing touches on my website/program this week. I say ‘website/program’ because what I am really doing is writing a program that produces a website. I am doing all the work in matlab and hardly writing any html. It would be best to have the program run automatically every day or two, but I’m not sure how to do this. I found out how to use colored scatter plots to give a big picture view of the float profiles. I was able to plot all profiles done by float 4738 using a colored scatter plot method. Here is an example of two of the graphs that will be on the site.

I will fix the dates and the positions of the color bar label next week. I like these graphs because they are very visual, these are an alternative to the three dimensional graphs I wanted to add last week. These graph plot three variable, just like the three dimensional graphs I wanted. I have also added links at the bottom of the page that allow visitors to look at plots of individual profiles. I have incorporated this all into my website program. Float 4738 is the only float that requires updating. However, I was initially building a program to update information on three floats. I spent a good amount of time revamping the program to run more efficiently. Originally every time the program updated the website, it started from scratch. It would process information from the first profile to the most recent. I have modified it so it saves past data in .mat an .fig files so when new profiles are available, it appends them onto the existing data. This makes much more sense and allows the program to run faster.

Float 6959 is no longer sending data, however it is sending its location. So I took it upon myself to write up a function to pull the location out of the partial files it sending. I have made 6959’s location part of the site updating program. So, even though we are not getting data we can still keep tabs on its location.

I will most likely include data from float 0005 and 0015 on the site, but have not added them yet. I also don’t have the routines for reading float 6810’s data. All three of these floats should be easy to add to the site after I compile the data. None of these floats will require to be updated by the program, so I don’t have to write any matlab code for these floats.

Next week I will polish up the website and update the paper with new information. Hopefully I will be able to collect some data in the lab next week as well. Our light source should be here next week; we can look at the reflected spectra of the light in the lab. Also, I am curious about the fluorescence function on the DiveSpec. I said in the paper that the spike at 550 could have been caused the LEDs of the DiveSpec. I am not sure how the blue light in the DiveSpec is produced. Is there a filter over white LEDs that only allows blue light through? If so, it’s possible that this does not stop ALL of the white light and there is still enough to see a small peak at 550. That is why I am hesitant to say that the peak at 550 is caused by fluorescence. Next week in the lab I will take a fluorescence spectrum of something that we know does not fluoresce and see if the peak still appears. Farther down the road, I look forward to going back to DMC to collect data!

See you next week

It’s finally Friday! This week I continued work on the float website. I organized most of my matlab scripts and nested them all into one script so I can run it and update all the pages. I am a bit stuck right now because I’m not sure how to present the data. Currently I have plotted all the profiles, so you get an idea of the general trend as depth increases. Some of the plots that Emmanuel’s scripts produce are logarithmic and others are not. I’ll need an explanation of why some are and some are not on a log scale. I would really like to use three dimensional graphs to plot variables such as temperature, salinity, and depth. What I want to do is make a three dimensional graph that visitors can move and spin with their mouse. This would be much better than having a couple jpeg pictures of a three dimensional graph. I worked hard on trying to do this, however have not found a way yet. I tried using vrml first. Matlab has a function that will export a .wrl file, which was very convenient. I was able to embed this file into html code after exporting it. However, vrml files require special downloads that allow you to see them, and it just did not look very good once on the webpage. It would not plot tick marks on the axis and would not rotate the graph correctly, so I bagged that idea.


Next I found a function someone wrote to put a three dimensional graph into java format and embed that in html. So I fooled around with this, but I ended up needing the symbolic math toolbox for matlab. If you know where I could get this toolbox it would be a great help. If I were to get this toolbox I could put graphs such as these on the site

http://wwwpub.zih.tu-dresden.de/~s9034647/peaksurface.html

This would allow me to use some pretty cool graphing techniques to show float profiles. Using java was the only way I found to put a decent looking 3D object on the site. I felt like learning java would not be the best use of my time, so I am holding out hope I can get my hands on the symbolic math toolbox. If not, less exciting 2D plots are always an option. Other than that, I have changed the way matlab finds the pixels needed for clickable areas on the stereographical map. I have made each float (4738,6959,6810) on the map appear as red, green, or blue. Then I read the image into matlab in RGB format and used that to locate the red, green, and blue pixels that correspond to a specific float. Finally I exported a second map that is presented on the website, with the floats appearing as uniform colors. On a side note, I am afraid 4738 is the only float that is sending usable data.

The float data and website building has kept me occupied most of the week, but I have also worked on the paper discussing the radiometer data we collected. This is done for the most part, however I didn’t know if was supposed to include DiveSpec data in this paper as well….? If so I could do that quite easily. I divided the samples by the closes spectralon plaque readings, which gave me a percent reflectance. What I wasn’t thinking about was that the spectralon plaque is only 95% not 100% reflectance. Dividing the sample by the spectralon plaque implies that plaque reflectance is 100%. This will cause the percent reflectance to be slightly higher than it truly is, but it will not change the shape of the graph. Correct me if I am wrong, but I don’t believe this to be a problem as long as it is mentioned in the paper. It is also unfortunate that the only way I can identify the algae and sponge readings is by “a species of sponge” or “a species of algae” and not a species name. It’s rather unscientific to say that a comparison was done between Didemnum sp. and a sponge. I am disappointed I couldn’t have been there to get a look at these sponges and get somewhat of identification. I just found out this week that my surgeon apparently never returned the doctor in Bangor’s call. So I need to get a hold of him and have him call this guy in Bangor. I am rather sour about this whole situation. I’ve been diving almost every weekend since the beginning of June. It’s unfortunate I can’t help with the in water radiometer work. In any case I want to have this cleared up before I head to DMC this fall.

I have been using the subplot tool in matlab frequently this week, both for the radiometer data and the float data. For example:

Also this week, Alina and I finally got some more pictures taken. We ended up putting the laser pointer as close to the camera as possible, but out of the frame. The solutions were made in five liters of water, and the pictures were taken in the same five liter bucket they were made in. The pictures came out relatively well; unfortunately I don’t have them on my computer to show you. They will require a fare amount of cropping to remove bits of reflected light off the bucket and surrounding objects. I’m glad we were able to get something usable though.

Looking toward the future, when Emmanuel gets back, I had some ideas about the next time we collect data for the Didemnum project. I have been thinking about it this week, and I think we should invest in a light source for the next time we take data. We should be collecting data in conditions similar to what will be present when the radiometer is on the AUV. So this means using the same light source that will be on the AUV. I have mentioned this light to Wayne and still think it would be a good choice to use as our illumination,

http://www.intova.net/products/intova-super-nova/

This is a power full LED light, that is very durable. It rated to 400 feet, although I’m not sure what kind of depth rating we are looking for. The beam angle is quite wide which I don’t see as a problem. Even if the acceptance angle of the radiometer is much smaller than the beam angle I still believe it would work fine. This would help insure that there would be even light coverage in uneven terrain. It would also make the mounting of the light on the AUV more flexible because it covers a wide area. Regardless I think we should purchase this light or another LED light that we can use next time we take samples. This way we have a consistent procedure for collecting the data, instead of using various dive lights. If it was the same light we plan to mount on the AUV that would be even better. Unfortunately we don’t have a waterproof xenon strobe, or a way to sync the strobe to the radiometer (that I know of). If we do want to take that route we will have to make a new plan where we can get some data using a strobe. Pretty much what I’m trying to say is we should keep the light source constant. The DiveSpec uses LEDs, our light on the AUV will be LEDs (I vote LEDs instead of the strobe), so we should use LEDs when we are collecting our data.

Last but not least I have contacted Chris down at DMC to get some information on the Didemnum growth down at DMC. He hasn’t been in around the dock lately but he saw quite a bit at Damariscove Island. He has referred me to Robert Russell at the Dept. of Marine Resources. Apparently he has added Didemnum to a survey of marine life along the coast of Maine. I’m sure he can clue us in on some good spots. That being said, it’s hard to beat the convenience of the dock at DMC and the nearby labs. The DiveSpec is self contained but the radiometer is not. I assume the radiometer can run off a battery, but the question is do we have such a battery? If we want to consider changing locations I will contact Mr. Russell.

That’s all that’s fit to print. Sorry if this was a bit lengthy.

Mapping!

This week has been a refreshing change of pace. I have enjoyed organizing the float data and thinking of creative ways to display it. I started this week by downloading some mapping software for matlab. It took me a couple hours to learn the ins and outs of the program. I am able to make some nice graphs using this software. These graphs can display the trajectory and current location of the floats. The software has a number of different style maps and a built in coastline generator. Here are a few Lambert maps I made displaying information on float trajectory.

I think information like this for each float would look nice on the website. This would be included along with graphs, tables, and raw data for people to download. I also mapped all the floats on a stereographic map.

 

The way I am envisioning it, this will be on the float homepage and each float (red dot) will be click-able, bringing the user to information about that float. I made the dots red on purpose, I will explain this later. I have started learning html in order to make the websites. I downloaded Microsoft Frontpage express, but have not used it much at all. I’m finding html not that difficult and it’s probably more beneficial to learn code rather than use Frontpage. So far it is going well. The commands are pretty strait forward, and as long I have a list of commands I can do pretty much whatever I need to.

After reviewing the University of Washington float page I got some ideas. It would be best is to have the website auto update. The University of Washington page updates itself with no human intervention. I believe I can make it so our float page requires no manual updating as well.

The first problem I ran into is much of html code is repetitive. Putting large amounts of data into a table using html requires you to type LOTS of code. So I used matlab to write code for me. This is the easiest way to make auto updating code. I can insert variables into a repetitive string, and then export it into a .htm file. This is good, because a matlab script can be used to write text to the .htm file that the site reads from, thus updating the information.

I spent a lot of time this week setting up the stereographic map above to update automatically. The way I have the map set up now, each red dot is a click-able region that opens a web page containing float information. However, when the float changes location, the map must update. Having the map itself update was quite simple, but in the html text the click-able region must be identified using pixels. So when the float moves location, the click-able regain must move as well. This meant correlating pixels with latitude and longitude. The way I solved this was to generate the map in matlab using the mapping software I downloaded, then exporting it out of matlab. Then I read it back into matlab as a grayscale image, and then used the intensity values to find the red dots. Red dots made this easy, because they are the only red pixels on the map. Finally, I wrote a string in matlab containing the html text and inserted the pixel locations into the string. Then I exported the string into an .htm file.

I don’t know how I am going to work everything out yet, but I got a pretty good start this week. I really like the idea of the website self updating. This may not be totally necessary, but I want to set it up this way. I believe the float information is sent to the ftp site that Emmanuel sent to me. I am banking on this, because my plan is to pull the data off that site (or any ftp site), then run some scripts and spit out html. However, I haven’t seen any updated information on the site since it was sent to me. Looking at past data, it appears that data sent almost daily. Also the more recent files from the float in the Atlantic contain no data.

Last but not least, I started writing up a report on the Didemnum spectra we took down at DMC in June. I didn’t spend too much time on this, but I got a start on it. One thing that would be good to include is what kind of dive light was used for the illuminated spectra. I’m guesses it was a halogen light but, I will have to double check. That about sums it up…

This week was spent mostly in the lab. I worked on finding the Cp* of bentonite clay and then taking pictures of back scatted light from solutions with different Cp values. To find the Cp* of the bentonite clay I used data collected by Wayne and Alina last week along with data I collected this week. I used the LISST to find the Cp of four different concentrations. The graph below shows the results.

I found the slope of the linear regression line formed by the points above. I forced the regression line through the origin of the graph because if there are zero particles, the Cp should be zero. The regression line is not on the graph above because I could not figure out how to force it through the origin in matlab. So I graphed it in excel to find the slope. I came up with a slope (Cp*) of 238936 mL/m*g or 0.24 m^2/g. Wayne's measurements after salt had been added to form aggregates was 0.4-0.5 m^2/g. So i believe my value to be reasonable and accurate enough for my purposes.

Next I attempted taking pictures using the mirror set up, having the laser shine through a hole in a mirror positioned at a 45 degree angle. Back scatted light should hit the mirror and be reflected to the camera lens. It seemed like i good idea, but i could not get it to work properly. The hole in the mirror was purposely smaller then the laser beam, this forced the laser beam to be a round beam. However this illuminated the inside of the hole in the mirror, the camera was picking up this light and ruining the picture. So i tried using a pinhole in a piece of paper to shrink the laser beam before it went through the mirror. This reduced the light to almost nothing. Next i tried drilling the hole in the mirror larger and making the pinhole larger. This worked to some extent, but the reflected light off the surface was still illuminating the hole in the mirror. After trying to get everything positioned just right for hours, I abandon the mirror idea.

Next up was our other idea of putting the laser pointer in the center of the camera frame. This way the camera looked down on the solution with the laser pointing directly in the center of the picture. The light reflected off the waters surface should be reflected directly back at the laser pointer and never make it to the camera lens. I tried using the polarized lens with this setup and found that it was eliminating the back scattered light. I went into a completely dark room to take the pictures and coated the inside of a beaker with black rubber to reduce reflection off the sides. I tried using a pinhole again but it made the pictures much darker and placed more unwanted objects in the picture. The pictures looked much better without it. Next I made the solutions and took some pictures. On the camera I used an iso of 1600 and an exposure time of 13 seconds. For the 0.5 1/m solutions I used pure water and could not see any back scattered light. For the rest i got reasonable pictures. Here are the pictures, from top to bottom the C total values are: 2 1/m, 8 1/m, and 32 1/m.

I am still kicking myself for the first picture. It looked great on the camera screen, but once i got it on the computer, i realized i must have moved it during the 13 second picture. I don't have a picture for pure water because the camera picked up very little light and the pictures looked terrible. The only thing i don't like about the above pictures is the ring formed at the edge of the beaker. I need a larger beaker that the edges will be out of the camera frame, or a lens that can focus over smaller distances. Given more time I don't see why I won't be able to get excellent pictures now that I have experimented.

Don't expect a post next week, because I will be relaxing and BBQing in the mountains of Vermont!